Domain Adaptation on VisDA synthetic-to-real
79.8Closed-set AccuracyPGPR
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| PGPRBackbone=DINOv2 (dinov2_vitl14), Encoder=Frozen, Classifier=Learnable2024.03 | 79.8 | 76.5 | 93.7 | 82.6 | 83.2 | |
| S+TBackbone=DINOv2 (dinov2_vitl14), Encoder=Frozen, Classifier=Learnable2024.03 | 78 | 73.2 | 91 | 79.9 | 80.5 | |
| UniOTBackbone=DINOv2 (dinov2_vitl14), Encoder=Frozen, Classifier=Learnable2024.03 | 77.9 | 68.7 | 87.6 | 78.4 | 78.2 | |
| DANCEBackbone=DINOv2 (dinov2_vitl14), Encoder=Frozen, Classifier=Learnable2024.03 | 75.1 | 67 | 93.6 | 78.9 | 78.6 |